MultiAxis Indexing Best Practices

[gakoenig]

Hi Guys-

So, I have parts that typically require some level of milti-axis indexing - typically for a hole to be drilled/tapped or a top profile to be cut. What I like to do is rough the part and cut a feature or two, index it on it's side for the drilling/profile, and roll it back over to finish off.

For the sake of accuracy, I use two WCSs in two separate operations (one for the primary work, a second for all the side work). Tool Orientation would be awesome, but not quite as accurate for now.

Trouble is, setting this all up isn't elegant. I basically wind up with each index of the part being a separate setup, leading to 3+ setups that all need to be managed.

Can I use Tool Orientation along with an independent WCS for the side operations? I always thought this was discouraged. Any better ideas here?

[LibertyMachine]

Are you talking about indexing with a 4/5 axis setup or manually flipping parts in the vise?

Do you have a sample file you could share so we have a better grasp of what you are up against?

[gakoenig]

I'm indexing with a 4 axis setup, with the rotary as the A axis.

Attached is an example file, the Setup 1, 2, and 3 are effectively Op1 and I post them together to run as one file. I would love to be able to program all this as one setup instead of breaking it all up like this.

The real and true proper way to do this is to dial in the rotary alignment, program off of the centerline, and use tool orientation. Having said that, indexing the side as it's own WCS seems way more accurate.

Thanks!

[LibertyMachine · Accepted solution]

So here's the thing:

Do you have a machine with TCP or some other form of dynamic work offsets? If so, you can dump all your operations into one Setup, turn on Tool Orientation where needed and everything is going to work just fine.

Don't have TCP or dynamic? You will have to make separate setups. Why?

With a stock post processor, tool orientation will only rotate the axis, but ti will do it around the original WCS. You can't define a new location on the part and have it pull a G55 XYZ value from that point.

Now, there is an edit that can be done to the post processor to enable a floating WCS that you can specify during tool orientation. There is still one problem, and that is the post processor is still going to output G54 throughout the program. So you are going to have G54 A0, A90 etc, all with different WCS positions but all the same WCS callout. You would need to perform a CTRL+H and go through the program and edit all the rogue offsets.

The problem with breaking them all out into separate Setups is that tool ordering is lackluster. If you have the same tools being run at separate times or multiple times, it's not going to be efficient. You would basically need to resign yourself to running one side clear through, then the next, so on and so forth.

I've argued this issue with the devs to no avail. I was hoping I could get them to see the limitations to their approach, but over the last 3 years, I can count on one hand the number of people who have come to the forums looking to do this. So i guess I'm in the minority. Sad, it's a very accurate (and quick) method of dealing with parts and fixtures that don't conform to theoretical perfect numbers.

[gakoenig]

Yea, I really wish this would get some development attention for this kinda stuff. Fusion has a lot of overlap with small scale manufacturers running slower machines that really benefit from cramming as much work on the table as possible and single-piece-flow. The WCS limitations make that a bit of a challenge unless you're going to fixture plate all the things. Even then, it can be frustrating.

Optimal would be using folders. Setup is where all the parts, stock, and fixtures would be laid out. Folders for individual operations on those parts with their own WCS. As far as safety is concerned, just withdraw Z to home when transitioning from one WCS to another.

I have noticed that Order by Tool seems to be working like one would expect again. I had a part on the 4th for the primary ops and the second ops in the vise - output it all in one post with Order by Tool and it worked perfectly (which I wasn't expecting). 

[sunderlandjoe]

What are you gaining by having a different WCS for every orientation?

Typically I'll just assign my rotary axis to say G59 and then set the G59 coordinates for Z and Y to the center of the rotary table. Align my part in CAM so my X lines up to wherever I want it. and use the tool orientation in fusion to machine all the sides in one setup.

The only time I would be worried about this is for some reason my rotary tables indexing was wrong, or it wasn't rotating around center (chuck not dialed in, or vice off center maybe?)

[LibertyMachine]

Trunnion tables on a 4th axis

Vise gets put on roughly in the same spot, part is loaded roughly where it should be. Change the parallels, change the vise, load the part slightly to one side..

All of those are valid reasons to have multiple work offsets.

Making a part out of bar stock rotating on centerline? Sure, one work offset works just fine, and I do that quite often.

Have a part and vise assembly that will not always be in the same spot? Multiple offsets.

[gakoenig]

If you want to be really accurate in production volumes with 4th axis parts, the standard is to use a different WCS for each index. Pretty much every production knife/tool shop here in Portland is on Speedios/Robodrills with direct drive rotary tables, and they all run 3-4 WCS setups for each part in this fashion.

Some of them run a direct WCS on each index (i.e. the WCS point they choose is directly on the part or fixture at that index), but it's also pretty common to use the rotary center XYZ and copy it 2-4 times to G55, G56, G57 etc and use each respective WCS to tweak things. In these environments, rotary tables get knocked around, stuff thermally grows in funky ways, folks might not have the fixture dialed in perfectly, etc. The kinematics of a 4th axis table are never as solid as those on a dedicated HMC or the rotary setup on a 5 axis machine, so if you're just doing index work, it's just easier to run multiple work coordinates.